AUTHOR=Xia Mengdi , Yu Ruiran , Zheng Zaiqiong , Li Huan , Feng Jie , Xie Xisheng , Chen Dongming 

TITLE=Meta-Analytical Accuracy of ANCA Renal Risk Score for Prediction of Renal Outcome in Patients With ANCA-Associated Glomerulonephritis

JOURNAL=Frontiers in Medicine

VOLUME=Volume 8 - 2021

YEAR=2022

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2021.736754

DOI=10.3389/fmed.2021.736754

ISSN=2296-858X

ABSTRACT=Abstract
Background: To evaluate the accuracy the diagnostic accuracy of ANCA renal risk score (ARRS) for prediction of renal outcome in ANCA-GN patients. 
Methods: We searched PubMed, EMBASE, Ovid, Web of Science, the Cochrane Library and ClinicalTrials.gov for studies which used ARRS to predict ESRD in patients with ANCA-GN. Two reviewers independently screened articles for inclusion, assessed the quality of studies with both an adapted QUADAS-2 tool. We calculated the combined ESRD patients in the ARRS categories and presented the summary and individual estimates based on the ARRS categories. Then the sensitivity, specificity, diagnostic odds ratio, positive/negative likelihood ratio and the area under the receiver operating characteristic (AUROC) curves of the pooled data for ARRS were used to assess the accuracy of the “above the low-risk threshold” (ARRS ≥ 2) and “high-risk grade” (ARRS ≥ 8) for renal outcome of ANCA-GN patients. Hierarchical summary receiver operating curves (HSROC) was used to verify the accuracy value. The clinical utility of ARRS was evaluated by Fagan plot.. 
Results: Twelve distinct cohorts from eleven articles involving 1568 patients with ANCA-GN were analyzed. The cumulative ESRD patients at the maximum follow-up of 60 months was 5% (95%CI [confidence interval]: 0.02 to 0.07; P < 0.001) for ANCA-GN with low ARRS (0-1 points) and significantly increased to 22% (95% CI: 0.15 to 0.29; P < 0.001) medium ARRS (2-7 points). The combined cumulative ESRD patients was 59% (95% CI: 0.49 to 0.69; P < 0.001) high ARRS (8-11points). The pooled sensitivity of ARRS ≥ 2 in predicting ESRD was 98% with a specificity of 30% and a DOR of 12.54, the mean AUROC value was 0.82. The pooled sensitivity of ARRS ≥ 8 in predicting ESRD was 58% with a specificity of 86% and a DOR of 7.59. The meta-regression and subgroup analysis indicated that variation in the geographic regions, study design, index risk, follow-up time, patient age, publication year, and patient number could be the potential sources of heterogeneity in the diagnosis of ARRS ≥ 8. 
Conclusion: This meta-analysis emphasized the good performance of the ARRS score in predicting the renal outcome in ANCA-GN.